Background and aims: Many sustaining traumatic brain injury (TBI) suffer ongoing post-concussion symptoms (PCS). The Rivermead Post-Concussion Symptoms Questionnaire (RPQ) is widely used, although there is disagreement about its structure. This study compared the fit of published RPQ structures with a four-factor structure derived from a large adult sample with TBI in Tasmania. Method: 661 adults with TBI completed the RPQ at approximately one month post injury. Exploratory factor analysis (EFA), using the first half of the sample (n = 330), suggested a four-factor solution. This was compared with models reported in the literature with the second half of the sample (n = 331), using structural equation modelling. Trajectory of recovery across these factors was examined within subsamples at 1, 3, 6 and 12 months following TBI. Results: Inter-correlations between items were strongest for somatic, cognitive and emotional functioning items and the EFA identified a four-factor model. Fit was examined utilising bootstrapping for model comparison. The data at 1 month following TBI best fitted the four-factor model (CFI = .95, RMSEA = .060 (.049–.071) and factors had adequate internal consistency (r = .61–.89). This model appeared a good fit and clinically useful across time points to 12 months post injury. Conclusions: Data best fitted a four-factor model, identified using a rigorous statistical approach. Clinicians and clinical researchers may use this preferred model to provide more specific measurement of the severity of PCS. Future research could attempt replication within international samples.